Tensorboard记录Loss变化

利用Tensorboard记录Loss值

运行环境：
tensorflow==1.14.0
keras==2.2.5

构建模型

CNN = build_cnn()
# type(CNN) = keras.engine.training.Model

log记录函数

def write_log(callback, names, logs, batch_no):
    for name, value in zip(names, logs):
        summary = tf.Summary()
        summary_value = summary.value.add()
        summary_value.simple_value = value
        summary_value.tag = name
        callback.writer.add_summary(summary, batch_no)
        callback.writer.flush()

创建Tensorboard记录

# log_path = './logs'
callback = TensorBoard(log_path)
callback.set_model(CNN)

在训练过程中记录Loss

datasize = X.shape[0]
batches = datasize//batch_size
for epoch in range(epochs):
	for batch in range(batches):
		loss = CNN.train_on_batch(x, y)
		write_log(callback, 
                  CNN.metrics_names,
                  loss,
                  epoch * batches + batch)